Deep OC-SORT: Multi-Pedestrian Tracking by Adaptive Re-Identification

Maggiolino, Gerard; Ahmad, Adnan; Cao, Jinkun

doi:10.48550/arxiv.2302.11813

Cited by 17 publications

(30 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Bot-SORT [1] introduces a ReID branch to extract appearance features corresponding to detections, selecting the smaller value between appearance cost and motion cost as the matching cost for detections and trajectories. Deep OC-SORT [15] enhances the model's ability to handle complex scenarios by incorporating appearance features on the basis of OC-SORT.…”

Section: Association Methodsmentioning

confidence: 99%

UniTracker:Transformer Based CrossUnihead for Multi-object Tracking

Wu,

Zhang

2024

Preprint

View full text Add to dashboard Cite

In recent years, tracking-by-detection (TBD) has emerged as the predominant approach for Multi-object Tracking (MOT). Most TBD algorithms typically employ separate branch heads to handle the coarse feature representations extracted from the backbone network. However, this approach often leads to suboptimal model performance. This is due to the presence of feature conflicts among multiple tasks on one hand, and the lack of information interaction among these tasks on the other. Inspired by CSTrack and Unihead, we have combined the feature decoupling module REN with the transformer-based task interaction module CIT to form CrossUnihead which is a plug and play module. This module not only effectively achieves task-based feature decoupling, but also facilitates information interaction among different tasks. The MOT algorithm based on CrossUnihead, named UniTracker, demonstrates impressive performance on the MOT dataset when compared to other advanced methods and is capable of achieving real-time tracking.

show abstract

Section: Association Methodsmentioning

confidence: 99%

UniTracker:Transformer Based CrossUnihead for Multi-object Tracking

Wu,

Zhang

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…In contrast, algorithms that integrate robust detection mechanisms with simplistic motion associations are witnessing continuous evolution. Despite progress in object detection and on-going optimization, many state-of-the-art end-to-end MOT models are still outperformed by algorithms blending motion or appearance association with potent detectors [28][29][30], even as newer methods employ intricate architectures and appearance embedding updates for performance enhancement. StrongSORT [28] addresses missing associations and detections in MOT by enhancing object detection, feature embedding, and trajectory association, incorporating a non-appearance information association model (AFLink), and Gaussian smooth interpolation using Gaussian Process Regression.…”

Section: Detection-based Multi-object Trackingmentioning

confidence: 99%

“…ByteTrack [29] mitigates the disruption from the loss of real objects and trajectory fragmentation by associating almost all detection boxes, recovering real objects, and filtering the background based on box and trajectory similarities. Deep OC-SORT [30] boosts the resilience of high-performance motion-based MOT methodologies against feature degradation by adaptively integrating appearance matching and leveraging object appearance information.…”

Section: Detection-based Multi-object Trackingmentioning

confidence: 99%

Infrared multi‐target detection and tracking in dense urban traffic scenes

Zha,

Luo,

2024

IET Image Processing

View full text Add to dashboard Cite

Infrared object detection and tracking in dense urban traffic remain a challenge due to factors such as low contrast, small intra‐class differences, and frequent false positives and negatives. To overcome these, the authors introduce YOLO‐IR, an algorithm based on the enhanced YOLOv8s, and YOLO‐DeepOC‐IR, a comprehensive infrared multi‐object tracking method for urban traffic, integrating both detection and tracking. During preprocessing, three infrared image enhancement techniques, local contrast multi‐scale enhancement, non‐local means, and contrast limited adaptive histogram equalization, are applied for better reliability in dense scenes. To further improve the performance, the original YOLOv8s backbone is replaced with MobileVITv3 to enhance detection accuracy and robustness. This infrared feature extraction module, incorporated into the detector, combines canny edge detection, Gabor filtering, and open operation layers, significantly boosting object detection in infrared imagery. The tracker's feature processing capabilities are improved using the learned arrangements of three patch codes descriptor and locality‐sensitive hashing for feature extraction and matching. Experimental results on FLIR ADAS v2 and InfiRay datasets indicate superior performance of this method, achieving 78.6% mAP and 151.1 FPS in detection, and up to 80.8% moving object tracking accuracy, 78.6% identification F1 score, and 62.1% higher order tracking accuracy in multi‐object tracking.

show abstract

“…Recent works such as OC-SORT [4] have made changes to SORT to handle non-linear motions, while others such as CBIoU [52] use a buffered version of IoU to extend bounding boxes for better object matching. More recently, reidentification-based trackers, like StrongSORT [12], have emerged, which focus on extracting more discriminative features for object re-identification to improve tracking results [2,29,57]. Some works have also been specifically developed for tracking team sport players [16,22,30,32,33].…”

Section: Multiple Object Trackingmentioning

confidence: 99%

Multi-task Learning for Joint Re-identification, Team Affiliation, and Role Classification for Sports Visual Tracking

Mansourian,

Somers,

De Vleeschouwer

et al. 2023

Proceedings of the 6th International Workshop on Multimedia Content Analysis in Sports

View full text Add to dashboard Cite

Effective tracking and re-identification of players is essential for analyzing soccer videos. But, it is a challenging task due to the nonlinear motion of players, the similarity in appearance of players from the same team, and frequent occlusions. Therefore, the ability to extract meaningful embeddings to represent players is crucial in developing an effective tracking and re-identification system. In this paper, a multi-purpose part-based person representation method, called PRTreID, is proposed that performs three tasks of role classification, team affiliation, and re-identification, simultaneously. In contrast to available literature, a single network is trained with multi-task supervision to solve all three tasks, jointly. The proposed joint method is computationally efficient due to the shared backbone. Also, the multi-task learning leads to richer and more discriminative representations, as demonstrated by both quantitative and qualitative results. To demonstrate the effectiveness of PRTreID, it is integrated with a state-of-the-art tracking method, using a part-based post-processing module to handle long-term tracking. The proposed tracking method, outperforms all existing tracking methods on the challenging SoccerNet tracking dataset. CCS CONCEPTS• Computing methodologies → Tracking.

show abstract

Deep OC-SORT: Multi-Pedestrian Tracking by Adaptive Re-Identification

Cited by 17 publications

References 0 publications

UniTracker:Transformer Based CrossUnihead for Multi-object Tracking

UniTracker:Transformer Based CrossUnihead for Multi-object Tracking

Infrared multi‐target detection and tracking in dense urban traffic scenes

Multi-task Learning for Joint Re-identification, Team Affiliation, and Role Classification for Sports Visual Tracking

Contact Info

Product

Resources

About